AI-Powered Novel BioAgeSense Device for Age Profiling Using Saliva

Sambasiva Rao Madamanchi

Abstract: This study introduces BioAgeSense, an AI-powered, non-invasive diagnostic device designed to estimate biological age using salivary biomarkers. Saliva samples from 200 participants (aged 20?70) were analyzed for physiological indicators such as DNA methylation (Horvath and Hannum clocks), cortisol, DHEA, 8-OHdG, TERT, inflammatory proteins, lactoferrin, urea, and microbiome composition. Biological age, derived from epigenetic clocks, served as the target variable. Machine learning models like Random Forest, XGBoost, and Deep Neural Networks (DNN) were developed, with the DNN achieving the best performance (R? = 0.89, MAE = 2.5, RMSE = 3.2). XGBoost and Random Forest followed closely, with R? values of 0.88 and 0.86, respectively. Participants were classified into low, moderate, and high-risk groups based on biological age acceleration scores. Key biomarkers related to stress, inflammation, and oxidative stress significantly influenced predictions. Distribution patterns of biomarkers included: cortisol (normal, ~5?ng/mL), DHEA (bell-shaped, ~2.5?ng/mL), 8-OHdG (right-skewed, ~10?ng/mL), Firmicutes/Bacteroidetes ratio (centered ~1.5), inflammatory proteins (~50?a. u.), lactoferrin (~4.5??g/mL), and urea (~28?mg/dL). BioAgeSense shows strong potential as a scalable, saliva-based platform for personalized aging assessment and remote health monitoring in precision medicine.

Keywords: BioAgeSense, AI-driven, Saliva, Novel, Aging, Machine learning